Preliminary Clusters Research Articles

Developing ensemble clustering through similarity measures: A semi‐supervised hierarchical clustering learning

SummaryA generic variation of hierarchical clustering (HC) that builds clusters from the bottom up is called agglomerative hierarchical clustering (AHC). The extension of AHC techniques using similarity criteria is the main topic of this research. Based on this, we create an AHC method that accomplishes clustering through ensemble approaches and combines the clustering of clusters with an original similarity measure. Three steps make up the proposed method's primary section. The first phase combines several individual AHC techniques to identify links between samples and create preliminary clusters. A heuristic similarity measure based on the developed clusters is used to determine how similar the samples are. The initial clusters produced using various techniques are all re‐clustered to create superclusters in the second step. The third phase involves creating the final clusters by assigning each sample to a supercluster with the greatest similarity after the clusters have been formed. Based on several benchmark datasets from the UCI machine learning repository, extensive experimental research has been done to assess the performance of the suggested approach. The outcomes unequivocally demonstrate that the suggested AHC‐based paradigm outperforms cutting‐edge techniques.

Improving unsupervised image clustering with spatial consistency

Unsupervised image clustering (UIC) is regularly employed to group images without manual annotation. One significant problem that occurs in the UIC context is that the visual-feature similarity across different semantic classes tends to introduce instance-dependent errors to clustering. The most successful recent approaches aimed at resolving this problem have focused on semisupervised reclustering, which utilizes reliable samples selected from the existing clusters. Despite this, virtually no previous work has considered the spatial consistency of the instance- and class-level representations which is crucial for error disambiguation. This makes it difficult to assess whether the selected reliable sample is reasonable. Accordingly, we propose a spatial consistency-based clustering (SCC) method to retain the alignment of representations learned from the instance and class levels in order to effectively select reliable samples from pre-existing clusters with errors thus leading to better clustering performance. More specifically, we first learn instance- and class-level representations by encouraging the semantic invariants of different instance augmentations and enforcing class alignment across semantically similar instances, respectively. We then assign instances with similar class-level representations to the same cluster to obtain the preliminary clusters. Subsequently we assess sample reliability by utilizing the spatial consistency constraint, which diffuses the class-level representations within the instance-level representation space. Finally, we employ semisupervised baselines combined with refinement techniques to perform reclustering based on the selected reliable samples. Extensive experimental results demonstrate that SCC can effectively obtain credible samples and outperform current SOTA clustering methods on the CIFAR-10 and CIFAR-100-20 benchmarks. The relevant code is available at https://github.com/RyanZhaoIc/SCC.git.

Effect of Play2Sleep on mother-reported and father-reported infant sleep: a sequential explanatory mixed-methods study of a randomized controlled trial.

This study evaluated the effect on infant sleep of a novel intervention (Play2Sleep) that combined infant sleep information with self-modeled video feedback on parent-infant interactions. An explanatory sequential mixed-methods design consisting of a randomized controlled trial with 63 mother-father-infant triads randomized to Play2Sleep or comparison home visit interventions was used. We used repeated measures analysis of covariance to detect changes in infant night wakings, nocturnal wakefulness, and sleep durations and Wilcoxon signed rank test to evaluate changes in perception of infant sleep problems. Family interviews (n = 20) were used to explain the quantitative findings and analyzed qualitatively using thematic analysis. Play2Sleep was effective in reducing maternal-reported infant wakefulness, F(1,55) = 5.33, P = .03, partial η2 = .09, and the number of paternal-reported naps, F(1,58) = 4.90, P = .03, partial η2 = .08. Parents in the Play2Sleep group reported significant improvements in problematic infant sleep that were not observed in the comparison group; however, Play2Sleep was not effective in reducing the number of parent-reported night wakings. Information overwhelm, learning infant cues, and working together with a subtheme of father involvement were key qualitative themes developed to explain the quantitative results. Unplanned exploratory analyses revealed a significant improvement in maternal depression symptoms in the Play2Sleep group. This study suggests Play2Sleep could improve infant sleep by promoting parental awareness of infant cues and father involvement and improving maternal depression. Additional research is needed to determine the optimal number and timing of sessions. Registry: ClinicalTrials.gov; Name: Play2Sleep: Using Play to Improve Sleep; URL: https://clinicaltrials.gov/ct2/show/NCT02742155; Identifier: NCT02742155. Keys EM, Benzies KM, Kirk VG, Duffett-Leger L. Effect of Play2Sleep on mother-reported and father-reported infant sleep: a sequential explanatory mixed-methods study of a randomized controlled trial. J Clin Sleep Med. 2022;18(2):439-452.

Beyond the hubble sequence – exploring galaxy morphology with unsupervised machine learning

ABSTRACT We explore unsupervised machine learning for galaxy morphology analyses using a combination of feature extraction with a vector-quantized variational autoencoder (VQ-VAE) and hierarchical clustering (HC). We propose a new methodology that includes: (1) consideration of the clustering performance simultaneously when learning features from images; (2) allowing for various distance thresholds within the HC algorithm; (3) using the galaxy orientation to determine the number of clusters. This set-up provides 27 clusters created with this unsupervised learning that we show are well separated based on galaxy shape and structure (e.g. Sérsic index, concentration, asymmetry, Gini coefficient). These resulting clusters also correlate well with physical properties such as the colour–magnitude diagram, and span the range of scaling relations such as mass versus size amongst the different machine-defined clusters. When we merge these multiple clusters into two large preliminary clusters to provide a binary classification, an accuracy of $\sim 87{{\ \rm per\ cent}}$ is reached using an imbalanced data set, matching real galaxy distributions, which includes 22.7 per cent early-type galaxies and 77.3 per cent late-type galaxies. Comparing the given clusters with classic Hubble types (ellipticals, lenticulars, early spirals, late spirals, and irregulars), we show that there is an intrinsic vagueness in visual classification systems, in particular galaxies with transitional features such as lenticulars and early spirals. Based on this, the main result in this work is not how well our unsupervised method matches visual classifications and physical properties, but that the method provides an independent classification that may be more physically meaningful than any visually based ones.

Refinement and Validation of the Head and Neck Lymphedema and Fibrosis Symptom Inventory

Lymphedema and fibrosis (LEF) are common yet overlooked late effects of head and neck cancer and its therapy. Lack of reliable and valid measures of head and neck LEF is a critical barrier to the timely identification and management of head and neck LEF. To fill this gap, we developed and pilot tested a 64-item patient-reported outcome measure ( Lymphedema Symptom Intensity and Distress Survey-Head and Neck, LSIDS-H&N). This article aims to report the process of further validation and refinement of the tool. A prospective, longitudinal study was conducted, and 120 patients with oral cavity and oropharyngeal cancer were recruited. Participants completed the LSIDS-H&N at pretreatment, end of treatment, and every 3 months up to 12 months after treatment. SAS PROC VARCLUS was used to generate preliminary clusters of item responses. Internal consistency of the item responses within each cluster was assessed using Cronbach's alpha. A total of 117 patients completed the study. The participants reported that the LSIDS-H&N was easy to understand and captured their symptoms and medical conditions. However, >50% of participants indicated that the survey was burdensome due to length. Thus, we proceeded with item reduction, and the shortened tool (33-item) was named Head and Neck Lymphedema and Fibrosis Symptom Inventory (HN-LEF Symptom Inventory). The subsequent exploration of symptom clusters identified 7 symptom domain clusters (eg, soft tissue and neurologic toxicity), all of which demonstrated good internal consistency. The HN-LEF Symptom Inventory has been carefully developed and refined to allow clinicians and researchers to capture LEF-associated symptom burden and function impairments. Additional rigorous psychometric testing of the tool is ongoing to further validate the strength and internal validity of this tool.

Urban Heat Island intensity and buildings’ energy needs in Duran, Ecuador: Simulation studies and proposal of mitigation strategies

Urban Heat Island (UHI) is a phenomenon that is affecting cities across the world. In tropical climates, an increased UHI intensity can cause a significant rise in heat stress probability, both indoors and outdoors. A general increase in the air-conditioning energy consumption of buildings should also be expected as a consequence. This paper evaluates the case of Duran, Ecuador. It estimated, through a spatial distribution analysis, urban morphology parameters, properties of materials, and anthropogenic heat emissions in the streets. Twenty-eight urban, randomly selected samples were classified using the k-means clustering technique, which resulted in four preliminary clusters. Exploratory UHI simulations were carried out in the Urban Weather Generator tool (UWG) using the four clusters to investigate the relevance of the main UHI driving factors through a sensitivity analysis. Urban samples were weighted, considering the sensitivity analysis, for better classifying the driving factors. Results indicate that Duran city is strongly affected by UHI, especially in informal settlements with high anthropogenic heat release. The urban temperatures obtained by the model were compared to the experimental measurements in different locations in the city. Besides, the building cooling needs were simulated by using the TRNSYS v. 17 tool for residential and commercial buildings of the city, showing that UHI is increasing cooling-related energy consumption in a 30–70 % for residential buildings and a 10–20 % for commercial buildings. Finally, general mitigation strategies and policy implications to reduce the UHI intensity are proposed, and a set of scenarios is tested for the cluster with the highest UHI intensity.

Preliminary clusters analysis based on functional disability scales for chronic Low Back Pain. Kinematic and EMG differences

Chronic Low Back Pain (cLBP) can be caused by several origins with important consequences on life, which induce psychosocial consequences. cLBP is not a pathology but a symptom. Patients with cLBP ...

Energy-Efficient Clustering Using Correlation and Random Update Based on Data Change Rate for Wireless Sensor Networks

This paper presents an energy-efficient clustering method using random update (EECRU) for wireless sensor networks. We divide the network into preliminary clusters depending on the temporal and spatial correlation of sensor data. These clusters are then updated by applying a dynamic update policy and a cluster head rotation scheme, where the change of sensor data and the energy residue of sensors are taken into consideration to achieve high energy efficiency and balance. The larger the change rate of the sensor data in a cluster, the higher its update frequency. A sensor node that has a high data frequency is more likely to be selected as the head of its cluster. In the data transmission phase, a sampling rate control method is adopted to improve the efficiency of data sampling. The experimental results indicate that the proposed EECRU method achieves high energy efficiency and energy balance.

Genome announcement: complete genome sequence of a novel Mupapillomavirus, HPV204

Human papillomaviruses (HPVs) are small, non-enveloped viruses with a circular double-stranded DNA genome, etiologically associated with various benign and malignant neoplasms of the skin and mucosa. As of May 30, 2015, 201 different HPV types had been completely sequenced and officially recognized and divided into five PV-genera: Alpha-, Beta-, Gamma-, Mu-, and Nupapillomavirus. The Mupapillomavirus genus currently consists of only two HPV types: HPV1 and HPV63, identified in 1980 and 1993, respectively, both associated with sporadic cases of cutaneous warts. In this preliminary study, we announce the complete genome sequence of a novel HPV type, now officially recognized as HPV204. Based on preliminary data, the genome of HPV204 comprises a total of 7,227 bp and contains five early open reading frames (E1, E2, E4, E6, and E7) and two late ORFs (L1 and L2). No E5 ORF could be identified. Preliminary HPV204 clusters to the Mu-PV genus, species Mu-3.

Structural Transitions of NiMn Precipitation Nanostructures as Determined by X-ray Diffraction Line Broadening and Mössbauer Spectroscopy

Structural transitions of nanometer-sized NiMn preprecipitates during isothermal aging of Fe-Ni-Mn alloys have been faintly indicated by the measurement of hardness and electrical resistivity changes. It is shown that a precise measurement of X-ray diffraction (XRD) line breadths in association with Mossbauer spectroscopy aims a better understanding of the structural transition of the precipitation nanostructures. Changes in the XRD line breadths clearly identified three stages corresponding to the stimulated clustering of solute elements, the formation of Guinier–Preston (GP) zones and subsequent transformation to the second-phase precipitates. Effective hardening takes place when preliminary solute clusters evolve to paramagnetic GP zones.

Sea SAR Images Analysis to Detect Oil Slicks in Algerian Coasts

In this paper, we investigate the performance of partition features derived from histogram analysis to isolate dark spots which are candidates to be oil spills in SAR images. The first partition is carried out to obtain preliminary clusters of the pixels on the basis of their grey level intensities and threshold values deduced from the histogram. The detection process is achieved by a contextual partition where the conflict pixels are attributed to their region involving local information about pre-etiqueted pixels neighbouring the pixel in question. For pixel’s assignment, we propose two decision criteria: the first based on Local Probability Maximization (LPM) while the second uses a Chi-squared test (χ2). We considered variable context in order to characterize the sea texture and dark spots. This method is tested on ERS-2 SAR Precision Image (PRI) covering Algerian coasts and gave promising results which are useful for the identification process.

The Trafficking in Persons Report: a game of risk

The State Department ranks countries on adherence to minimum standards set forth by the Trafficking Victims Protection Act 2000. The Trafficking in Persons Report (TIP) is updated annually and failure to enact changes to combat trafficking results in higher tier rankings. This paper evaluates the TIP by situating this tool in light of special features of the modern era, such as globalization and risk. Through a survey of the theoretical literature on risk and on trafficking risk factors, we devise six preliminary risk clusters and discuss how the TIP could incorporate governments’ response to trafficking risk factors into the ranking system. Our intentions are to spark debate about how risk factors could be incorporated in the TIP, to provide a preliminary model and to encourage further research in this area.

Porcine SWC1 is CD52—Final determination by the use of a retroviral cDNA expression library

During the last decades for several species –e.g. swine – many mAb to leukocyte-specific molecules have been developed and clusters of differentiation corresponding to human CD could be established. However, for a significant amount of the raised mAb the corresponding antigens were not characterized on the molecular level and therefore preliminary clusters – in swine so-called Swine workshop clusters (SWC) – were established. These clusters contain antigens with currently no obvious orthologs to human leukocyte differentiation antigens. In this study, we describe the generation of a eukaryotic cDNA expression library from in vitro activated porcine peripheral blood mononuclear cells. Screening of this library with an antibody recognizing SWC1 enabled isolation and sequencing of cDNAs coding for the porcine SWC1 molecule. A BLAST search of the obtained sequence revealed that SWC1 is the orthologous molecule of human CD52. Therefore, our study provides the basis for comparative studies on the role of CD52 in different mammalian species. In addition, the established cDNA library can be used for investigation of additional SWC-defined molecules.

PIRSF: family classification system at the Protein Information Resource.

The Protein Information Resource (PIR) is an integrated public resource of protein informatics. To facilitate the sensible propagation and standardization of protein annotation and the systematic detection of annotation errors, PIR has extended its superfamily concept and developed the SuperFamily (PIRSF) classification system. Based on the evolutionary relationships of whole proteins, this classification system allows annotation of both specific biological and generic biochemical functions. The system adopts a network structure for protein classification from superfamily to subfamily levels. Protein family members are homologous (sharing common ancestry) and homeomorphic (sharing full-length sequence similarity with common domain architecture). The PIRSF database consists of two data sets, preliminary clusters and curated families. The curated families include family name, protein membership, parent-child relationship, domain architecture, and optional description and bibliography. PIRSF is accessible from the website at http://pir.georgetown.edu/pirsf/ for report retrieval and sequence classification. The report presents family annotation, membership statistics, cross-references to other databases, graphical display of domain architecture, and links to multiple sequence alignments and phylogenetic trees for curated families. PIRSF can be utilized to analyze phylogenetic profiles, to reveal functional convergence and divergence, and to identify interesting relationships between homeomorphic families, domains and structural classes.

Summary of workshop findings for porcine T-lymphocyte-specific monoclonal antibodies

Fifty-seven monoclonal antibodies (mAb) selected after the first round analyses in the Third International Swine CD workshop for their possible reactivity with T-lymphocyte specific antigens were further analysed in a second round. As target cells for flow cytometric analyses served peripheral blood mononuclear cells, nylon-wool enriched T-lymphocytes, thymocytes, splenocytes, and lymphocytes derived from Peyer’s patches. These second round analyses revealed 15 different data sets. Together with 22 pre-selected data sets from the first round analyses with the whole panel of monoclonal antibodies, 37 data sets were used for the clustering of the respective mAb. Using the LTDB4 program, 19 preliminary clusters could be defined. Two clusters ( C3 and C7) with 4 mAb showed no labelling of resting T-lymphocytes. Seven clusters ( C1, C2, C4, C5, C6, C11, and C12) contain mAb (in total: 16 mAb) directed against subsets of CD4 −CD8 − T-lymphocytes. These mAb seem to recognise antigens on porcine T-lymphocytes with T-cell receptor (TcR) γ/δ chains. Three clusters ( C8, C9, C10, C13) seem to be artificial. They contain either mAb staining CD4 −CD8 − T-lymphocytes and low CD8+ cells ( C8, C9), mAb with various reactivity ( C10) and mAb with known differences in their reactivity (C13). Cluster C14 contains 3 mAb against the CD4a-epitope, C15 describes mAb directed against porcine CD8c-epitope whereas mAb against CD8a and CD8b-epitopes grouped in C19. The mAb found in C16 seem to recognise CD45R. Cluster C17 is composed of different standards directed against CD2, CD3, CD5 and wCD6. Two additional mAb recognising the CD2a-epitope could be enclosed. C18 contains two mAb directed against SWC2.

Biochemical analysis of preliminary clusters in the non-lineage panel

After the preliminary clustering of the whole workshop panel using flowcytometry data, we selected those clusters that were composed of mAbs that bound a broad range of cells. To obtain evidence that mAbs in each of these preliminary clusters detected the same antigen, they were tested for their capacity to compete with each other for binding to a target cell and the molecular weights of their antigens were estimated after immunoprecipitation. Most preliminary clusters in the non-lineage panel contained control mAbs that had been characterised in one of the previous workshops, and this study therefore increased the number of mAbs available to each of these non-lineage antigens. One new interesting cluster, BoWC14, was described which defined an antigen on myeloid cells and a subpopulation of CD4, CD8 and WC1 T lymphocytes (BT3/8.12 and IL-A155). An additional cluster, which did not fit into any other panel, contained mAbs specific for an antigen restricted to the erythroid lineage and received the nomenclature BoWC15 (ANA8, IL-A135, IL-A137, IL-A138 and IL-A160).

Analysis of bovine B cell reactive monoclonal antibodies

Monoclonal antibodies (mAbs) submitted to the workshop B cell panel were screened for reactivity with bovine surface immunoglobulin positive (SIg +) cells from tonsil, mesenteric lymph nodes (MLN), ileal Peyer's patches (IPP), thymus and peripheral blood by fluorescence activated cell sorter (FACS) analysis. All mAbs within B cell panel reacted with B cells except the negative control mAb VPM61. Although most mAbs stained the majority of B cells from different tissues, 14 mAbs did not stain B cells from IPP. Detailed FACS analysis of the 24 mAbs in preliminary clusters PC17, PC28, PC29, PC30, PC35, PC41 and PC43 showed all preliminary clusters (PC) to contain mAbs with similar and variable specificities. This absence of clear-cut clusters therefore did not permit a simple classification of B cell surface antigens via the B cell mAbs and suggests considerable complexity and variability of the B cell surface.

Reactivity of monoclonal antibodies with bovine blood mononuclear cells activated by mitogens and superantigens

Monoclonal antibodies forming seven preliminary clusters in the third ruminant CD workshop (PC15, 16b, 16c, 20b, 26b, 26c and 30) were analysed by flow cytometry upon reaction with in vitro stimulated bovine mononuclear cells. Concanavalin A (Con A), pokeweed mitogen (PWM) and three Staphylococcus aureus enterotoxins (superantigens SEA, SEB and SEC2) were used as stimulating agents. The mAbs in PC15, 16c, and 26b reacted uniformly irrespective of the stimulating agent, in contrast to clusters PC16b, 20b, 26c and 30, which could be subgrouped according to differential reaction pattern depending on mitogen or superantigen stimulation. The mAbs CC28 and CACT114A seem to detect an antigen preferably induced by superantigens. MAb GB16A was shown to react strongly with Con A. Depending on the activation antigen, its time course of expression was either the same upon superantigen stimulation as that for mitogen-stimulated cells or slightly delayed.

Ruminant cluster CD62L

Within the Third Workshop, four monoclonal antibodies (mAbs) were considered to belong to this cluster: IAH-CC32 (3W-360), BAQ92 (3W-485), Buf44 (3W-452) and IVA94 (3W-563). IAH-CC32 was the workshop control and has previously been identified as a CD62L-specific mAb (Howard, C.J., Sopp, P. and Parsons, K.R., 1992. L-selectin expression differentiates T cells isolated from different lymphoid tissues in cattle but does not correlate with memory. Immunol., 77:228–234) by showing that human L-selectin mAb LAM1.4 (Spertini, O., Kansas, G.S., Reimann, K.A., Mackay, C.R. and Tedder, T.F., 1991. Functional and evolutional characteristics of distinct epitopes on the leukocyte adhesion molecule-1 (TQ-1, Leu-8) that regulate leukocyte migration. J. Immunol., 147:942) cross-reacts with bovine cells and blocks binding of IAH-CC32. The mAb DREG-56 to human CD62L (Bosworth, 1993. Bovine L-selectin: a peripheral lymphocyte homing receptor. Vet., Immunol., Immunopathol, 37:201–215) also stains cattle cells. Buf44 blocked binding of IVA94 and IAH-CC32, indicating that all three bind to the same or closely related epitope on the L-selectin molecule (Naessens, 1996. Biochemical analysis of preliminary clusters in the non-lineage panel. Vet. Immunol. Immunopathol., 52: 347–356).

Cluster analysis

Cluster analysis was performed on flow cytometry data generated from the reactivities of the 302 workshop monoclonal antibodies with 36 target cell preparations. The antibodies were assigned to 42 preliminary clusters that were subjected to further examination in subsequent stages of the workshop.